CPAP as a useful tool in COVID-19 related Acute Hypoxemic Respiratory Failure: Experience from 3 UK Hospitals

Introduction: Severe acute hypoxemic respiratory failure (AHRF) in COVID-19 pneumonia is associated with a high mortality rate, resulting in mounting pressures on intensive care units worldwide. Different oxygenation management protocols are used in different centres. Most centres switch patients who fail to oxygenate adequately using conventional oxygen therapy (COT) methods to non-invasive positive pressure ventilation (NIPPV), usually continuous positive airway pressure (CPAP). Other centres resort to invasive mechanical ventilation (IMV) directly, without a trial of NIPPV. In this trial, we aim to compare the ecacy of different approaches in managing COVID-related AHRF, and ascertain if CPAP therapy reduces the need for IMV. Methods: We carried out a retrospective cohort study on patients with laboratory-conrmed COVID-19 at three university hospitals in Essex, United Kingdom. We included all patients with signicant AHRF (dened as needing oxygen therapy FiO 2 more than 0.4 to maintain an oxygen saturation of 92%) who were deemed eligible for IMV escalation during a 3-month period (1 st March to 31 st May 2020). Results: Out of 174 patients who met the criteria, 84 patients received CPAP (Group 1). Half needed intubation (n=42). 90 patients did not have a CPAP trial (Group 2). 76.6% needed intubation (n=69). No difference was found between the two groups in demographic criteria or disease severity. Our results show a signicant difference in 60-day mortality between group 1 and 2 (25% versus 37.8%, p=0.02). COT as standalone therapy for COVID-19 patients (group 2) was associated with a trend of more increased risk of intubation and an increased relative risk of death (RR 2.14, 95% CI 1.39 to 3.29). This corresponds to a number needed to treat (NNT) of 3.74 (95% CI 2.47 to 7.73). Patients in group 1 who failed CPAP trial and required intubation did not have an increased risk of mortality when compared to group 2 patients who required intubation. Conclusion: Our results support introducing CPAP rather than escalating FiO2 suggests CPAP can be safely used to treat patients with Clinical trials are needed to guide recommendations for optimum timing and selection of patients most likely to Data presented n (%) or mean±SD, Statistical signicance at results, COPD: chronic obstructive pulmonary disease, CKD: Chronic kidney disease, LoS: Length of stay.


Introduction
The COVID-19 pandemic has caused a huge rise in patients presenting with Acute Hypoxemic Respiratory Failure (AHRF) and mounting pressures on the Intensive Care Units (ICU) across the globe. There is a signi cant surge in reports from around the world, describing different oxygenation management protocols in different centres. However, there is still variation in the management of COVID-19 related AHRF as evidence by the heterogeneity of protocols in different centres, even within the same country.
Most centres switch patients who fail to oxygenate adequately using conventional oxygen therapy (COT) methods (e.g. Venturi Mask or Non Rebreathing Mask (NRM)) to Non-Invasive Positive Pressure Ventilation (NIPPV) usually Continuous Positive Airway Pressure (CPAP), High ow nasal Oxygen (HFNO), or in the minority of patients with CO 2 retention, Bi Level Positive Airway Pressure (BiPAP). Other centres resort to Invasive Mechanical Ventilation (IMV) directly without a trial of NIPPV. Reducing the risk of Aerosol Generated Procedures (AGP) related infection is one of the reasoning behind the latter approach.
Marini and Gattinoni strongly support early IMV to avoid progression of the mild disease to a more severe lung injury secondary to the spontaneous vigorous inspiratory efforts leading to patient self-induced lung injury (P-SILI). (1) Others argued that P-SILI should not be used as a justi cation for pre-emptive IMV, with all its documented complications, without giving the patient a trial of CPAP / BiPAP. (2) The timing of the endotracheal intubation (ETI) and the location for initiation of IMV vary widely. While some centres will have their severely hypoxemic patients moved to ICU, in others patients are intubated in wards prior to being moved to the ICU. Increasingly, new data to support the use of CPAP/ BiPAP prior to initiation of IMV is being published.
We present in this retrospective study the data collected from three University Hospitals in Essex, United Kingdom comparing different approaches of management of COVID related AHRF. Our aim was to ascertain if CPAP therapy would reduce the need for ETI.

Methods
Inclusion criteria: All adult patients (≥18 years) with con rmed laboratory COVID-19 and signi cant AHRF de ned as needing oxygen therapy (FiO 2 more than 0.4) to maintain an oxygen saturation of 92%, and who were deemed eligible for escalation for IMV during a 3-month period (1 st of March to 31 st of May 2020) from three University hospitals in UK. Patient data and outcomes were retrospectively extracted from patient records.
Data were collected retrospectively and anonymously and no patient consent was required.
Exclusion Criteria: Patients with limits of care which included do not intubate (DNI) order as per the treating physician.
Statistical analysis: Qualitative and quantitative variables were summarized with frequencies (absolute and relative, percentage) and central tendency (means and medians) and variability (standard deviations, SD, and interquartile ranges, IQR) indicators, depending on their parametric distribution. Student t test as well as Chi-Square test were used for quantitative and qualitative variables. Kaplan-Meier survival analysis using Log rank test compared 60-days mortality between the CPAP versus non-CPAP groups using R software ("survminer" package).

Results
Patients requiring CPAP were treated in either negative pressure or rooms with adequate air changes with non-invasive monitoring. Health care workers caring for patients used full personal protective equipment (PPE) as per Public Health England advice. Non vented masks were used. All patients were treated in the respiratory wards (acute respiratory care areas) and managed by the respiratory medical and nursing team with input from the critical care outreach team. CPAP was initiated at pressures of 5-8 cm H 2 O and increased up to 10-12 cm H 2 O depending on patient comfort and treatment response. In one of the three sites in the trust, CPAP was not used and patients who did not respond were triaged to IMV directly.
174 patients met the above inclusion criteria. They were subsequently divided into two groups: Group 1: Patients who received CPAP treatment (n=84) Group 2: Patients who did not have a CPAP trial (n=90) Clinical details and outcomes are shown in table 1. The outcome of the whole cohort of patients is shown in gure 1. There was a signi cant 60-days mortality difference between group 1 and 2 (25% versus 37.8%, p=0.02). Twenty-one patients in group 1 died, including one before invasive mechanical ventilation after suffering a pulmonary embolism.

Discussion
In this retrospective study, we compared COVID-19 patients who had CPAP prior to IMV with those who were directly treated with IMV after receiving only conventional oxygen therapy (COT). Mortality in patients who were invasively ventilated was not different in both groups (Table 1). However, the overall mortality was signi cantly less in the CPAP group (25% vs 37.8%, p=0.02). There was no baseline difference in the patient demographics and severity of illness. Our analysis shows that COT as standalone therapy for COVID-19 patients (group 2) was associated with a trend of more intubation a relative risk of death of 2.14 which was statistically signi cant (95 % CI 1.39 to 3.29). This corresponds to a number needed to expose (NNE) of 3.74 (95% CI 2.47 to 7.73) (i.e. 4 patients would have to receive CPAP (instead of only oxygen therapy) to reduce mortality by one).
CPAP trial failure is not associated with increased mortality when compared to patients who needed ETI without a CPAP trial as the mortality in both these groups who went on to have IMV was similar (Table 1).
Our study though small, suggests CPAP can be safely used to treat patients with AHRF and adds to a number of studies recently published with similar results. An Italian group was able to show in a prospective multicentric work (n=157) that helmet CPAP can help to avoid intubation in nearly half of COVID-19 patients (27). A single centre before-after French study (n=52) showed CPAP to be signi cantly associated with reduced need for intubation in COVID-19 patients (11).
Our results support introducing CPAP rather than escalating FiO2 in refractory to COT cases. The exposure to high FiO2 is known to be associated with harm in the form of impaired lung function and resorption atelectasis (22). Some subgroup of patients can bene t from CPAP and avoid ETI with all its associated adverse effects.
Frat et al compared NIV to COT and HFNO in AHRF and concluded no difference in the rate of ETI (13). The underlying mechanism can be the lung injurious vigorous spontaneous breathing, a phenomenon known as P-SILI (patient self-in icted lung injury). P-SILI is more probable when a vigorous respiratory effort is superimposed on severe ARDS adding insult to injury (9). On the other hand, an increase in transpulmonary pressure and tidal volume after CPAP can theoretically lead to a higher regional lung stress and strain especially in non-homogenous ARDS lungs (10). To note that both P-SILI and diaphragmatic injury are possible after prolonged and strenuous breathing even without any respiratory support (e.g. CPAP) (8,10). In fact, considering atelectasis as a risk factor for P-SILI, lung recruitment by CPAP, better oxygenation (hence reduced hypoxic drive) and more homogenous distribution of ventilation may reduce the risk of P-SILI in CPAP treated patients. (8,14,15) The added risk of infection transmission boosted the early views leaning to avoid CPAP in COVID-19 patients. While It is di cult to con rm the risk associated with every AGP, the role of NIV remains without strong evidence (16 Our studies have many limitations: It is a retrospective small sample study and not randomised. However, we think it can add along with other works to the cumulative knowledge about CARDS. We mainly used one CPAP interface (face mask) and cannot report on others especially Helmet which showed bene t in other works (26,27). Our work shows that CPAP trial is warranted in COVID-19 patients presenting with AHRF under close monitoring. This trial can be safely applied in out of ICU respiratory units by a trained team.
We also believe that CPAP could be a valuable tool in low resourced settings and also it is included in WHO COVID 19 clinical guidance (28). Clinical trials are needed to guide recommendations for optimum timing and selection of patients who most likely to bene t.
We still await the results of randomised control trial to compare CPAP, HFNO and standard oxygen or the IMV (29). Until such robust data are published, our observation suggest that CPAP has a role in management of type 1 respiratory failure due to COVID 19. Declarations